CrewAI
CrewAI is an open-source Python framework for building multi-agent systems based on the role/goal/backstory metaphor. Agents are defined by their function (role), success criterion (goal), and persona (backstory); tasks bundle a description with an expected output; crews assemble agents into sequential or hierarchical processes. Tool use, delegation, and memory are first-class features.
The framework is configurable via Python or YAML. CrewAI distinguishes itself from LangGraph by leading with a team metaphor rather than an explicit graph, which trades graph-level control for a more intuitive on-ramp for non-developers building their first multi-agent system.
Example
A market-research crew might pair a "Senior Research Analyst" agent (gather sources) with a "Strategy Writer" agent (synthesize findings) and a "Reviewer" agent (fact-check) — all running sequentially, each task's output flowing into the next. Switching to a hierarchical process introduces a manager LLM that decides which crew member handles each subtask, useful when the work has branching paths but expensive when it does not.
Frequently asked questions
What is CrewAI?
- CrewAI is an open-source Python framework for building multi-agent systems based on the role/goal/backstory metaphor.
How does CrewAI work?
- The framework is configurable via Python or YAML. CrewAI distinguishes itself from LangGraph by leading with a team metaphor rather than an explicit graph, which trades graph-level control for a more intuitive on-ramp for non-developers building their first multi-agent system.
Can you give an example of CrewAI?
- A market-research crew might pair a "Senior Research Analyst" agent (gather sources) with a "Strategy Writer" agent (synthesize findings) and a "Reviewer" agent (fact-check) — all running sequentially, each task's output flowing into the next. Switching to a hierarchical process introduces a manager LLM that decides which crew member handles each subtask, useful when the work has branching paths but expensive when it does not.